Minimally Invasive Glucose Monitoring Using a Highly Porous Gold Microneedles-Based Biosensor: Characterization and Application in Artiﬁcial Interstitial Fluid

: In this paper, we present the first highly porous gold (h-PG) microneedles-based second-generation biosensor for minimally invasive monitoring of glucose in artificial interstitial fluid (ISF). A highly porous microneedles-based electrode was prepared by a simple electrochemical self-templating method that involves two steps, gold electrodeposition and hydrogen bubbling at the electrode, which were realized by applying a potential of − 2 V versus a saturated calomel electrode (SCE). The highly porous gold surface of the microneedles was modified by immobilization of 6-(ferrocenyl)hexanethiol (FcSH) as a redox mediator and subsequently by immobilization of a flavin adenine dinucleotide glucose dehydrogenase (FAD-GDH) enzyme using a drop-casting method. The microneedles-based FcSH / FAD-GDH biosensor allows for the detection of glucose in artificial interstitial fluid with an extended linear range (0.1–10 mM), high sensitivity (50.86 µ A cm − 2 mM − 1 ), stability (20% signal loss after 30 days), selectivity (only ascorbic acid showed a response about 10% of glucose signal), and a short response time (3 s). These properties were favourably compared to other microneedles-based glucose biosensors reported in the literature. Finally, the microneedle-arrays-based second-generation biosensor for glucose detection was tested in artificial interstitial fluid opportunely spiked with different concentrations of glucose (simulating healthy physiological conditions while fasting and after lunch) and by placing the electrode into a simulated chitosan / agarose hydrogel skin model embedded in the artificial ISF (continuous glucose monitoring). The obtained current signals had a lag-time of about 2 min compared to the experiments in solution, but they fit perfectly into the linearity range of the biosensor (0.1–10 mM). These promising results show that the proposed h-PG microneedles-based sensor could be used as a wearable, disposable, user-friendly, and automated diagnostic tool for diabetes patients.


Introduction
Optimal diabetes management requires accurate real-time monitoring of glucose. People with diabetes have to test their blood sugar level several times every day to know how different foods, medications, and activities may affect it [1,2]. The normal blood glucose level (tested while fasting) for non-diabetics should be between 3.9 and 7.1 mmol/L (70-130 mg/dL). The mean normal blood glucose and HCl [39]. The templating method combines various techniques, such as the flow-stream technique, sputter deposition, and electrochemical methods, and involves the following three steps: preparation of the template, deposition of gold, and removal of the template. The self-templating method is a simplified and time-saving templating method that involves only two steps: gold electrodeposition and gas bubbling as a self-template [38,[40][41][42].
In this work, a highly porous gold (h-PG) microneedles-based biosensor for minimally invasive sensing of glucose in artificial interstitial fluid was realized and investigated. The h-PG surface was developed by electrodeposition of a gold layer followed by hydrogen bubbling at the electrode surface through applying a negative potential of −2 V versus a saturated calomel electrode (SCE) in a solution containing NH 4 Cl as a supporting electrolyte and a hydrogen source. The h-PG microneedles were characterized and compared to microneedles modified with multiwalled carbon nanotubes (MWCNTs) and to a planar gold electrode before and after the modification with h-PG and MWCNTs. The flavin adenine dinucleotide glucose dehydrogenase (FAD-GDH)/6-(ferrocenyl)hexanethiol (FcSH)/h-PG microneedles-based electrode was successfully realized by immobilization of the redox mediator 6-(ferrocenyl)hexanethiol and the FAD-GDH enzyme [43]. An image of the microneedles and a schematic representation of the as-assembled microneedles-based electrode are reported in Figure 1A,B, respectively. To the best of our knowledge, this is the first example of a h-PG-modified microneedles-based electrode reported in the literature. Finally, the feasibility of the developed biosensor for minimally invasive monitoring of glucose was evaluated in artificial interstitial fluid and in a simulated skin model.
Catalysts 2019, 9, x FOR PEER REVIEW 3 of 15 most active element in the alloy, which is gold, which could be done either electrochemically or chemically. Electrochemical deposition is an electrodissolution-disproportion-deposition reaction between gold and HCl [39]. The templating method combines various techniques, such as the flowstream technique, sputter deposition, and electrochemical methods, and involves the following three steps: preparation of the template, deposition of gold, and removal of the template. The selftemplating method is a simplified and time-saving templating method that involves only two steps: gold electrodeposition and gas bubbling as a self-template [38,[40][41][42].
In this work, a highly porous gold (h-PG) microneedles-based biosensor for minimally invasive sensing of glucose in artificial interstitial fluid was realized and investigated. The h-PG surface was developed by electrodeposition of a gold layer followed by hydrogen bubbling at the electrode surface through applying a negative potential of −2 V versus a saturated calomel electrode (SCE) in a solution containing NH4Cl as a supporting electrolyte and a hydrogen source. The h-PG microneedles were characterized and compared to microneedles modified with multiwalled carbon nanotubes (MWCNTs) and to a planar gold electrode before and after the modification with h-PG and MWCNTs. The flavin adenine dinucleotide glucose dehydrogenase (FAD-GDH)/6-(ferrocenyl)hexanethiol (FcSH)/h-PG microneedles-based electrode was successfully realized by immobilization of the redox mediator 6-(ferrocenyl)hexanethiol and the FAD-GDH enzyme [43]. An image of the microneedles and a schematic representation of the as-assembled microneedles-based electrode are reported in Figure 1A,B, respectively. To the best of our knowledge, this is the first example of a h-PG-modified microneedles-based electrode reported in the literature. Finally, the feasibility of the developed biosensor for minimally invasive monitoring of glucose was evaluated in artificial interstitial fluid and in a simulated skin model.  Figure  1A is reproduced from ref. [28] with permission of Elsevier Science ltd (2019).

SEM and Electrochemical Characterization of h-PG/Au Microneedles Electrodes
SEM experiments were carried out to investigate the morphology of the highly porous gold microneedles-based electrode. SEM images of the bare and the h-PG-modified microneedles-based electrode are presented in Figures 2A and 2B (mag. x100 and mag. x1000) and 2 C (mag. x1000), respectively. It is possible to note the well-defined pores with diameters of approximately 25 µ m, as shown in Figure 2C. The highly porous gold surface is probably obtained by the reduction of Au 3+  Figure 1A is reproduced from ref. [28] with permission of Elsevier Science ltd (2019).

SEM and Electrochemical Characterization of h-PG/Au Microneedles Electrodes
SEM experiments were carried out to investigate the morphology of the highly porous gold microneedles-based electrode. SEM images of the bare and the h-PG-modified microneedles-based electrode are presented in Figures 2A and 2B (mag. ×100 and mag. ×1000) and 2C (mag. ×1000), respectively. It is possible to note the well-defined pores with diameters of approximately 25 µm, as shown in Figure 2C. The highly porous gold surface is probably obtained by the reduction of Au 3+ ions to Au 0 atoms with their subsequent aggregation and deposition on the rough surface of the microneedles electrode to form the porous structure. A key point of the fabrication of the porous surface is the choice of a suitable negative potential and the right time at which to apply it. A negative Catalysts 2019, 9, 580 4 of 14 potential of −3 V generated hydrogen bubbling at the microneedles surface of the working electrode that was too vigorous and destroyed the gold surface. A lower potential (about 0 V) was not enough to generate the proper hydrogen bubbling to create the porous structure. The potential of −2 V was therefore chosen as a compromise to guarantee a sufficient bubbling of hydrogen at the gold surface to create the proper porous structure without ruining the surface itself. A time of 60 s was chosen as it is the length of time that enabled the porous structure to entirely cover the electrode. Another important parameter that was carefully optimized was the distance between the bottom of the microneedle arrays structure and the beaker's bottom. The highly porous gold surface was produced when this gap was not greater than 3 mm, probably because a larger distance would not facilitate the diffusion processes of H + , AuCl 4 , and Cl − ions towards and away from the electrode surface to enable the aggregation and deposition of the gold atoms on the electrode surface to form the porous structure [44]. ions to Au 0 atoms with their subsequent aggregation and deposition on the rough surface of the microneedles electrode to form the porous structure. A key point of the fabrication of the porous surface is the choice of a suitable negative potential and the right time at which to apply it. A negative potential of −3 V generated hydrogen bubbling at the microneedles surface of the working electrode that was too vigorous and destroyed the gold surface. A lower potential (about 0 V) was not enough to generate the proper hydrogen bubbling to create the porous structure. The potential of −2 V was therefore chosen as a compromise to guarantee a sufficient bubbling of hydrogen at the gold surface to create the proper porous structure without ruining the surface itself. A time of 60 s was chosen as it is the length of time that enabled the porous structure to entirely cover the electrode. Another important parameter that was carefully optimized was the distance between the bottom of the microneedle arrays structure and the beaker's bottom. The highly porous gold surface was produced when this gap was not greater than 3 mm, probably because a larger distance would not facilitate the diffusion processes of H + , AuCl4, and Cl − ions towards and away from the electrode surface to enable the aggregation and deposition of the gold atoms on the electrode surface to form the porous structure [44]. In order to demonstrate the improvements in the electrodeposition method in terms of electroactive area, the gold film was characterized electrochemically in 0.5 M H2SO4 at a scan rate of 100 mV s −1 . The cyclic voltammogram shows two anodic peaks due to gold oxidation starting at about 1 V versus SCE and one large cathodic peak at about 0.7 V versus SCE due to the reduction of the gold oxide formed during the positive sweep, as shown in Figure 3A. The real surface area of the highly porous gold electrode was calculated by integrating the charge required for reducing the gold oxide ( Figure 3A, red curve), and the results were compared with those obtained with a bare microneedles gold electrode with the same geometric surface ( Figure 3A, black curve). The h-PG microneedles electrode showed a real surface area of 88.50 cm 2 , assuming that the charge density for the reduction of gold oxide is 390 μC cm −2 , a value that is about 100 times higher compared to the bare microneedles gold electrode (0.90 ± 0.02 cm 2 ). In order to demonstrate the improvements in the electrodeposition method in terms of electroactive area, the gold film was characterized electrochemically in 0.5 M H 2 SO 4 at a scan rate of 100 mV s −1 . The cyclic voltammogram shows two anodic peaks due to gold oxidation starting at about 1 V versus SCE and one large cathodic peak at about 0.7 V versus SCE due to the reduction of the gold oxide formed during the positive sweep, as shown in Figure 3A. The real surface area of the highly porous gold electrode was calculated by integrating the charge required for reducing the gold oxide ( Figure 3A, red curve), and the results were compared with those obtained with a bare microneedles gold electrode with the same geometric surface ( Figure 3A, black curve). The h-PG microneedles electrode showed a real surface area of 88.50 cm 2 , assuming that the charge density for the reduction of gold oxide is 390 µC cm −2 , a value that is about 100 times higher compared to the bare microneedles gold electrode (0.90 ± 0.02 cm 2 ).
The real active area of the porous surface was also investigated by cyclic voltammetry in a solution of Fe(CN) 6 3− / 4− ( Figure 3B,C). A remarkable enhancement of the anodic and cathodic peak current densities after the h-PG electrodeposition was observed as a consequence of the increase in the electroactive surface area. By comparing these results with those obtained with a planar gold electrode before and after the h-PG electrodeposition, it can be easily noted that the increase in the current after the h-PG electrodeposition is larger in the case of the Au-microneedles electrode ( Figure 3B) compared to the Au-planar electrode ( Figure 3C). The roughness factors of the two electrodes before and after the electrodeposition of h-PG have been calculated, and the roughness factor after the electrodeposition of h-PG resulted to be almost 20 times higher in the case of the microneedles electrode (ρ = 1032.1 ± 2.3) compared to the planar electrode (ρ = 56.0 ± 0.8) (Table S1). The particular geometry of the microneedles probably allows for a better electrodeposition of h-PG and therefore a larger electroactive surface area is obtained. Table S1 also shows the electroactive area and roughness factor of the microneedles and planar gold electrodes modified by electrodeposition of Au-MWCNTs, according to a procedure reported in our previous work [29], for comparison. It is interesting to note the large enhancement of the roughness factor in the case of the h-PG microneedles-based electrode, which results to be three times higher than that obtained with microneedles modified with Au-MWCNTs ( Figure 3B,C, blue curves). The heterogeneous electron transfer rate constants (k 0 , cm s −1 ) for the planar gold and the microneedle gold electrodes before and after the modification with h-PG and Au-MWCNTs are also reported in Table S1, calculated with a method proposed by Lavagnini et al. that merges the Klingler-Kochi and Nicholson and Shain methods developed for irreversible and reversible systems, respectively [45,46]. The Au microneedles/h-PG electrode showed the highest k 0 value (k 0 = 56.2 ± 0.5 × 10 3 cm s −1 ), attesting to faster electron transfer kinetics. The real active area of the porous surface was also investigated by cyclic voltammetry in a solution of Fe(CN)6 3− / 4− (Figures 3B and 3C). A remarkable enhancement of the anodic and cathodic peak current densities after the h-PG electrodeposition was observed as a consequence of the increase in the electroactive surface area. By comparing these results with those obtained with a planar gold electrode before and after the h-PG electrodeposition, it can be easily noted that the increase in the current after the h-PG electrodeposition is larger in the case of the Au-microneedles electrode ( Figure  3B) compared to the Au-planar electrode ( Figure 3C). The roughness factors of the two electrodes before and after the electrodeposition of h-PG have been calculated, and the roughness factor after the electrodeposition of h-PG resulted to be almost 20 times higher in the case of the microneedles electrode (ρ = 1032.1 ± 2.3) compared to the planar electrode (ρ = 56.0 ± 0.8) (Table S1). The particular geometry of the microneedles probably allows for a better electrodeposition of h-PG and therefore a larger electroactive surface area is obtained. Table S1 also shows the electroactive area and roughness factor of the microneedles and planar gold electrodes modified by electrodeposition of Au-MWCNTs,

Electrochemical Characterization of FcSH/h-PG/Au Microneedles Electrodes
The h-PG/Au microneedles-based electrode was subsequently modified by deposition of the redox mediator 6-(ferrycenyl)hexanethiol (Fc-SH), by keeping the electrode overnight in a solution of 8 mM FcSH in phosphate buffer at pH 7.5. The voltammetric profile of the modified electrodes shows a couple of quasi reversible redox peaks, as shown in Figure 4, at different scan rates. It is interesting to note that the voltammograms show the typical shape of a surface-bound species with a small separation between the anodic and cathodic peaks. The formal redox potential of FcSH was determined to be about +0.38 V versus SCE, in good agreement with the value reported in the literature for this mediator [47]. An analysis of the Faradaic current as a function of v resulted in a linear relationship, as shown in the inset of Figure 4, attesting to the typical dependence of the current on the scan rate of a surface-confined mediator.
The effect of FcSH amount was investigated and the maximum current signal was obtained by using an 8 mM FcSH solution for the mediator immobilization (data not shown). Higher amounts of FcSH showed a decrease in current response, probably due to a steric hindrance effect of the mediator on the microneedles electrode surface [48,49].
The h-PG/Au microneedles-based electrode was subsequently modified by deposition of the redox mediator 6-(ferrycenyl)hexanethiol (Fc-SH), by keeping the electrode overnight in a solution of 8 mM FcSH in phosphate buffer at pH 7.5. The voltammetric profile of the modified electrodes shows a couple of quasi reversible redox peaks, as shown in Figure 4, at different scan rates. It is interesting to note that the voltammograms show the typical shape of a surface-bound species with a small separation between the anodic and cathodic peaks. The formal redox potential of FcSH was determined to be about +0.38 V versus SCE, in good agreement with the value reported in the literature for this mediator [47]. An analysis of the Faradaic current as a function of v resulted in a linear relationship, as shown in the inset of Figure 4, attesting to the typical dependence of the current on the scan rate of a surface-confined mediator.
The effect of FcSH amount was investigated and the maximum current signal was obtained by using an 8 mM FcSH solution for the mediator immobilization (data not shown). Higher amounts of FcSH showed a decrease in current response, probably due to a steric hindrance effect of the mediator on the microneedles electrode surface [48,49].

Application of the h-PG/Au Microneedles-Based Glucose Biosensor in Artificial ISF
Before studying the amperometric responses toward D-glucose of the assembled FAD-GDH/FcSH/h-PG/Au microneedles-based glucose biosensor, the amount of loaded enzyme was carefully examined and optimized by drop-casting onto the microneedles surface FAD-GDH amounts varying from 1 to 5 U. The variations in the current densities as a function of the enzyme loading are shown in Figure S1. The biosensor response increased proportionally with the enzyme loading up to 1.5 U, corresponding to 5 μl of enzyme. A slight decline in the response signal is observed at higher enzyme concentrations probably due to the rigid structure of the FcSH molecule,

Application of the h-PG/Au Microneedles-Based Glucose Biosensor in Artificial ISF
Before studying the amperometric responses toward D-glucose of the assembled FAD-GDH/FcSH/h-PG/Au microneedles-based glucose biosensor, the amount of loaded enzyme was carefully examined and optimized by drop-casting onto the microneedles surface FAD-GDH amounts varying from 1 to 5 U. The variations in the current densities as a function of the enzyme loading are shown in Figure S1. The biosensor response increased proportionally with the enzyme loading up to 1.5 U, corresponding to 5 µL of enzyme. A slight decline in the response signal is observed at higher enzyme concentrations probably due to the rigid structure of the FcSH molecule, showing a rate-limiting dependence on the enzyme concentration only at low loaded FAD-GDH amounts. Therefore, 1.5 U was the concentration of enzyme utilized for the biosensor's development. Figure 5A shows a comparative cyclic voltammogram of the FAD-GDH/FcSH/h-PG microneedles-based electrode in the absence and in the presence of glucose in artificial interstitial fluid. A good electrocatalytic current was registered with an onset potential of +35 mV, in agreement with the value reported in the literature for the redox potential of the FcSH mediator (E 0 = +38 mV vs. SCE).
In Figure 5B, it is possible to observe that the calibration curve for glucose detection in artificial interstitial fluid has an extended linear range between 0.1 and 10 mM glucose (inset Figure 5B) and a sensitivity of 50.86 µA cm −2 mM −1 (R 2 = 0.97, n = 5). The detection limit resulted to be 50 µM, calculated using the relation 3σ/S, with σ and S the absolute standard deviation of the intercept and the slope of the calibration curve, respectively. By fitting the calibration curve to the Michaelis-Menten kinetic parameters, a J max value of 520.20 ± 25.11 µA cm −2 was obtained and a Michaelis-Menten constant (K M ) of 11.75 ± 1.18 mM was calculated, showing a good affinity of the FAD-GDH enzyme for glucose. microneedles-based electrode in the absence and in the presence of glucose in artificial interstitial fluid. A good electrocatalytic current was registered with an onset potential of +35 mV, in agreement with the value reported in the literature for the redox potential of the FcSH mediator (E0 = +38 mV vs. SCE).
In Figure 5B, it is possible to observe that the calibration curve for glucose detection in artificial interstitial fluid has an extended linear range between 0.1 and 10 mM glucose (inset Figure 5B) and a sensitivity of 50.86 μA cm −2 mM −1 (R 2 = 0.97, n = 5). The detection limit resulted to be 50 μM, calculated using the relation 3σ/S, with σ and S the absolute standard deviation of the intercept and the slope of the calibration curve, respectively. By fitting the calibration curve to the Michaelis-Menten kinetic parameters, a Jmax value of 520.20 ± 25.11 μA cm −2 was obtained and a Michaelis-Menten constant (KM) of 11.75 ± 1.18 mM was calculated, showing a good affinity of the FAD-GDH enzyme for glucose.

Effect of pH and Temperature on the Stability and Selectivity of the Glucose Biosensor
The median pH and temperature values are important parameters for enzymes to maintain their activities and catalysis efficiency. The optimum pH and temperature values for the FAD-GDH enzyme in soluble form was found to be pH 7 and 35 °C, respectively [50]. A possible shift in optimal pH can be obtained for immobilized enzymes. The pH effect was studied by varying the pH of the

Effect of pH and Temperature on the Stability and Selectivity of the Glucose Biosensor
The median pH and temperature values are important parameters for enzymes to maintain their activities and catalysis efficiency. The optimum pH and temperature values for the FAD-GDH enzyme in soluble form was found to be pH 7 and 35 • C, respectively [50]. A possible shift in optimal pH can be obtained for immobilized enzymes. The pH effect was studied by varying the pH of the ISF in the range between 3 and 9, using acetate, phosphate, and TRIS buffer. The maximum current responses were obtained in phosphate buffer pH 7.5 ( Figure 6A) at 35 • C ( Figure 6B). These results are of great consequence for a potential application of the microneedles-based biosensor under human physiological conditions.
An important requirement for a microneedles-based biosensor for transdermal analysis is the ability to operate over several days with no significant decrease in its response. The stability was examined using a 0.2 mM glucose solution over 30 days by using the biosensor in continuous operation for 10 measurements every day. A high stable current response was obtained over the whole period, with a decrease in the initial signal response of about 20% after 30 days ( Figure 6C). Such a high lifetime of the biosensor probably reflects the stability of the enzyme in the highly porous structure of the microneedles electrode surface.
Another important feature of a microneedles-based biosensor is the capacity to avoid interference from other electroactive compounds present in the human interstitial fluid. The response of the proposed biosensor was evaluated with a standard glucose concentration and with equal concentrations of exogenous interfering compounds, such as d-fructose, d-mannitol, d-galactose, and ascorbic acid. These compounds have a similar oxidation potential to glucose and could affect the biosensor's current response. Figure 6D shows the high selectivity of the proposed biosensor, as no tested compound showed any significant current signal with the exception of ascorbic acid, whose signal was, however, about 10% of the signal obtained for glucose. Also, xylose and maltose were tested, but they did not show any current signal, confirming the high selectivity of the proposed biosensor.
are of great consequence for a potential application of the microneedles-based biosensor under human physiological conditions. An important requirement for a microneedles-based biosensor for transdermal analysis is the ability to operate over several days with no significant decrease in its response. The stability was examined using a 0.2 mM glucose solution over 30 days by using the biosensor in continuous operation for 10 measurements every day. A high stable current response was obtained over the whole period, with a decrease in the initial signal response of about 20% after 30 days ( Figure 6C). Such a high lifetime of the biosensor probably reflects the stability of the enzyme in the highly porous structure of the microneedles electrode surface.
Another important feature of a microneedles-based biosensor is the capacity to avoid interference from other electroactive compounds present in the human interstitial fluid. The response of the proposed biosensor was evaluated with a standard glucose concentration and with equal concentrations of exogenous interfering compounds, such as D-fructose, D-mannitol, D-galactose, and ascorbic acid. These compounds have a similar oxidation potential to glucose and could affect the biosensor's current response. Figure 6D shows the high selectivity of the proposed biosensor, as no tested compound showed any significant current signal with the exception of ascorbic acid, whose signal was, however, about 10% of the signal obtained for glucose. Also, xylose and maltose were tested, but they did not show any current signal, confirming the high selectivity of the proposed biosensor.  An analytical comparison with the results obtained with other microneedles-based glucose biosensors recently reported in the literature is shown in Tables 1 and 2. It is interesting to note that most microneedles-based biosensors show extended linear ranges and high sensitivities; however, they are based on the oxidation of H 2 O 2 (first-generation biosensors), which occurs at high overpotentials with the obvious drawback of the possibility of side reactions, such as the oxidation of ascorbic acid, lactic acid, and uric acid, thus hampering the selectivity of the biosensor [51][52][53][54][55]. The second-generation FAD-GDH/FcSH/h-PG/Au microneedles-based biosensor proposed in this study exhibits a linear range Catalysts 2019, 9, 580 9 of 14 and a sensitivity comparable to the devices based on the "first generation" biosensor mechanism without the drawback of the high overpotential and the dependence on oxygen concentration, extremely important requirements for a subcutaneous biosensor, while maintaining at the same time a much larger linear range compared to the second-generation microneedles-based biosensor based on MWCNTs reported in the literature [23,56]. Table 1. Heterogeneous electron transfer rate constant (k 0 ), real electroactive area (A EA ), and roughnesss factor (ρ) of the gold planar electrode and the gold microneedles-based electrode before and after the electrodeposition of h-PG and Au-MWCNTs. A geo planar Au electrode = 0.020 cm 2 ; A geo microneedle electrode = 0.2 cm 2 .

Real ApplicAtion
Ref.

Simulation of Continuous Glucose Monitoring Using a Skin Model
The feasibility of the FcSH/h-PG/Au microneedles-based biosensor for continuous glucose monitoring was tested in flow conditions for 12 h through the determination of the glucose concentrations in a skin model, consisting of a chitosan/agarose hydrogel opportunely embedded in artificial ISF at two different glucose concentrations, 4 mM and 8 mM, in order to simulate the normal glucose levels in a healthy patient under fasting conditions (in the morning before breakfast) and after lunch (approximately 6 h after breakfast). Both current signals presented a lag-time of about 2 min compared to the experiments performed in solution, possibly because of the time of diffusion of glucose through the hydrogel; the signals resulted to be quite stable, thus attesting to the robustness of the developed device. The current signal registered for the 8 mM glucose concentration resulted to be about 2 times higher than that registered at the 4 mM glucose concentration, confirming that the two concentrations fall perfectly into the linearity range of the glucose biosensor (0.1-10 mM). FAD-glucose dehydrogenase (FAD-GDH) from Aspergillus niger was obtained from Creative Enzyme Ltd. The FAD-GDH (activity 300 U/mL) was solubilized in a 10 mM PBS buffer (pH 7.5) and stored at −20 • C until further use.

Modification of Au Microneedles-Based Electrode
The microneedles-array Au microneedles electrodes were kindly provided by a collaborator at Glasgow University [25,43]. These electrodes are based on a polycarbonate scaffold (0.5 × 0.5 × 0.02 cm) with 64 microneedles divided as four 4 × 4 arrays. Each pyramid showed the following dimensions: base 0.06 cm; height 0.1 cm; 4 × 4 array area 0.2 cm 2 . On this platform, three electrodes were used as working electrodes (gold), while the fourth electrode was the reference electrode (silver). However, in the present work, we used an external saturated calomel electrode as a reference electrode to run the preliminary experiments.
At the beginning, the microneedles electrodes were electrochemically cleaned by sweeping the potential between −0.3 and 1.7 V versus SCE 20 times at a scan rate of 300 mV s −1 in 0.5 M H 2 SO 4 . The electrodeposition of highly porous gold (h-PG) was performed by cycling the potential between +0.8 and 0 V versus SCE for 25 scans at 50 mV s −1 in a solution containing 10 mM HAuCl 3 and 2.5 M NH 4 Cl. Afterwards, a fixed potential of −2 V versus SCE was applied in the same solution in order to activate the self-templated formation of pores. Next, the electrodes were activated in 0.5 M H 2 SO 4 by running CVs between 0 and +1.7 V versus SCE at a scan rate of 100 mV s −1 for 25 cycles. All the electrochemical parameters (e.g., electroactive surface area, heterogeneous electron transfer rate constant (k 0 , cm s −1 ), and roughness factor (ρ)) were obtained in 5 mM Fe(CN) 6 3−/4− (50 mM PBS buffer pH 7.5 + 137 mM NaCl). Furthermore, the as-prepared h-PG/Au microneedles electrode was modified through self-assembling monolayer formation by incubating the electrode in an 8 mM FcSH solution overnight. After thoroughly washing the electrode with working buffer, 5 µL of FAD-GDH enzyme were cast onto the modified electrode.

Modification of Au Planar Electrode
Planar gold electrodes (AuE) (d = 1.6 mm, AMEL, Milano, ITALY) were carefully cleaned with Piranha solution (1:3 mixture of conc. H 2 O 2 with H 2 SO 4 , CAUTION: Piranha solution is especially dangerous, corrosive and may explode if contained in a closed vessel, it should be handled with special care). Afterwards, they were mechanically cleaned with polishing cloths and deagglomerated alumina slurry of 1 µm in diameter (SIEM, Bologna, ITALY), and successively sonicated in ultrapure water for 5 min. Next, the electrodes were electrochemically cleaned by sweeping the potential in the range between −0.3 and 1.7 V versus SCE at a scan rate of 300 mV s −1 M for 20 cycles in 0.5 M H 2 SO 4 . The h-PG surface was realized by a self-templating method, as reported above. The as-modified electrodes were characterized and modified following the microneedles protocol reported in the previous section.

Conclusions
In this work, we described the fabrication and characterization of a novel glucose biosensor based on microneedles technology. The novelty of the work consists in the realization of a h-PG gold microneedles electrode and in the use of the mediator 6-(ferrocenyl)hexanethiol, thus realizing a second-generation biosensor with improved performance in terms of linear range, sensitivity, and selectivity. The highly porous film was electrochemically fabricated through a deposition/self-templating method that made the active surface area of the microneedles electrode up to 100 times larger than that of a classical bare one, thus enhancing the sensitivity of the biosensor. The device was tested in artificial interstitial fluid, showing good repeatability and good long-term stability, and also under simulated continuous monitoring conditions in a chitosan/agarose hydrogel skin model. The obtained results are encouraging for the use of the proposed microneedles-based biosensor as a wearable device to be used in medicine for the continuous monitoring of glucose in patients with diabetes. Investigation of in vivo performance through experiments in healthy and diabetic volunteers will be the next aim of the work. Further work will be devoted also to the realization of toxicity tests of the developed device, on the one hand, as well as to studies of a possible integration of suitable electronics for wireless communication, on the other hand.